using-hackinator
Skill1 starsChanged 3 months agoMITView on GitHub
Use when the user asks which Hackinator mode or skill fits, wants orientation to the hacker ethic, or has not yet distinguished building from refining. Not for a concrete new build or prototype — that's builder — or polishing working code — that's whittler
In this repository23 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .cursor/rules/beads.mdcCursor rule
- beadsSkill
- builderSkill
- consulting-codexSkill
- dispatching-subagentsSkill
- evidence-before-claimsSkill
- finishingSkill
- hacking-workflowSkill
- naming-as-designSkill
- planningSkill
- reading-like-proseSkill
- reviewing-workSkill
- shapingSkill
- spike-and-stabilizeSkill
- subtraction-firstSkill
- test-firstSkill
- using-beadsSkill
- using-worktreesSkill
- walking-skeletonSkill
- whittlerSkill
- writing-skillsSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 29
Two heads beat one on hard problems: When you're stuck or the stakes are high, consult an independent model (consulting-codex) — then judge …
- Distinctive: line 30
Play: If it isn't a little bit fun, you're doing it wrong.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Using Hackinator
Overview
Hacking in the original sense: not breaking in, but building for the joy of it and making the result beautiful. A hack is a clever, playful, well-crafted solution. This suite has two moods and the techniques behind them.
Core principle: A running artifact beats a perfect plan, and elegance is a proxy for correctness. Build first, then reveal the simple thing hiding inside what you built.
The Ethic (what these skills assume)
Skills advise how to work. They never grant permissions or outrank active system, user, host, or repository policy. Apply every technique inside those boundaries.
- Hands-on imperative. You understand a system by building with it, not by reading about it.
- Beauty is functional. Elegant code is easier to trust, change, and love. Ugliness is a smell.
- Subtraction is progress. The best hack removes more than it adds.
- Ship something that runs. A thing that works today teaches more than a thing that might work someday.
- Adapt to the host. Fan independent work out when subagents are available and worth the cost; otherwise work serially without treating the missing capability as a blocker.
- Track it in the open. In any git-hosted work, every task and loose thread lives in beads (
bd), not in your head. A thought you don't file is a thought you'll lose. - Two heads beat one on hard problems. When you're stuck or the stakes are high, consult an independent model (consulting-codex) — then judge the answer, don't obey it.
- Play. If it isn't a little bit fun, you're doing it wrong.
The Workflow (the front door for any non-trivial build)
For real work — anything past a quick edit — start with hacking-workflow. It sequences the whole loop and right-sizes decisions and ceremony to the work (a spike skips most of it):
shape → isolate → plan → build → craft → verify → review → finish| Phase | Skill | In one line |
| Shape | shaping | Resolve intent and consequential ambiguity before code |
| Isolate | using-worktrees | Isolate edits when policy, capability, and collision risk call for it |
| Plan | planning | Map files + right-sized tasks, filed as beads |
| Build | builder / test-first | Make it work; keeper code is test-first |
| Craft | whittler | Make it elegant, behavior-preserving |
| Verify | evidence-before-claims | No "done" without fresh command output |
| Review | reviewing-work | Use the best available independent check, with a self-review fallback |
| Finish | finishing | Follow the authorized landing choice; clean up |
Track the whole thing in beads throughout (using-beads).
Two Personas
| You want to... | Invoke | Mood |
| Make an idea real, fast — prototype, spike, scratch an itch | builder | Make it work |
| Take working code and make it elegant, minimal, a joy to read | whittler | Make it beautiful |
Personas are entry points. They lean on the shared techniques below and will pull them in as needed.
The Techniques (used by both personas, or directly)
| Skill | Use when |
| walking-skeleton | Starting something new — get a thin end-to-end slice running before filling anything in |
| spike-and-stabilize | Facing an unknown — throwaway exploration to retire risk, then rebuild for keeps |
| subtraction-first | Improving code — the instinct to add when removing is the better move |
| naming-as-design | Something feels tangled — names that hide the design vs. names that reveal it |
| reading-like-prose | Making code a joy to read — structuring for the human who reads it next |
| dispatching-subagents | Splitting work across permitted subagents or parallel calls when doing so materially helps |
| using-worktrees | Isolating parallel or risky edits when the host and repository permit it |
| using-beads | Tracking work in bd — always, in any git-hosted activity: tasks, deferred TODOs, discovered bugs, dependencies |
| consulting-codex | Getting an independent second opinion, adversarial gut-check, or outside diff review from the Codex CLI when stuck or the stakes are high |
How to Choose
Non-trivial build or change? → hacking-workflow (runs the authority-aware loop)
Nothing runs yet? → builder (which starts with walking-skeleton)
It runs but it's ugly? → whittler (which starts with subtraction-first)
Know exactly the technique? → invoke the technique/process skill directlyAuthoring New Skills
To add or edit a hackinator skill, use writing-skills (watch it fail without the skill first, write the minimal skill, close loopholes). Keep the flat namespace, active names, and "Use when..." descriptions that state triggers only — never a workflow summary.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

